基于无人飞机成像的桥梁裂缝形状和宽度识别研究

发布时间：2018-08-21 12:18

【摘要】：桥梁作为重要关键节点,承担着日益增长的交通压力,桥梁技术状况直接关系交通和人身安全,进行桥梁技术状况评定尤为重要。混凝土桥梁病害的最大表征就是开裂,以往采用人工借助桥检车读取裂缝宽度,难以准确描述裂缝扩展情况。以无人飞机成像获得桥梁表面原始图像,进行分析处理,形成裂缝扩展信息图,既避免了人工精度误差又提高了检测速度,同时不受地形桥型桥宽的限制,有着重要的研究价值。本文研究了无人飞机桥梁检测中成像角度修正问题,提出以三点激光器获得三点物距,推导平均物距、被测平面相对夹角,对物距法获得的像素解析度进行修正。针对提出的无人飞机桥梁检测成像特点,基于形态学的组合滤波方法构建了图像预处理算法,增强图像对比度,保护边缘信息。对无人飞机成像图像进行边缘识别,提出与Otsu相结合的Canny边缘检测算法,通过形态学判据消除微小噪点,得到完整裂缝形态。针对裂缝形态提出基于最小二乘法拟合裂缝中心线的法向宽度识别算法,并考虑特殊情况裂缝形态的识别分析,定义交叉裂缝交叉区域宽度。获得具有工程实际意义的法向裂缝宽度,为基于无人飞机成像的桥梁技术状况评定提供数据支持。本文最后进行实桥检测对比实验,对湘潭市湘江二大桥进行无人飞机成像裂缝识别,将宽度识别结果与第三方检测机构检测数据进行对比,结果表明无人飞机成像裂缝识别具有良好精度,并优于人工桥梁裂缝检测形成裂缝扩展信息图;满足工程实际需要,具有工程应用价值和发展潜力。
[Abstract]:As an important key node, bridge bears increasing traffic pressure. The bridge technical condition is directly related to traffic and personal safety, so it is very important to evaluate the bridge technical condition. Crack is the biggest representation of concrete bridge disease. In the past, it is difficult to accurately describe the crack propagation by using manual aid of bridge inspection vehicle to read the crack width. The original image of bridge surface was obtained by unmanned aerial imaging, and the crack spreading information map was formed by analyzing and processing, which not only avoided the error of artificial precision, but also improved the detection speed, and was not limited by the width of topographic bridge at the same time. It has important research value. In this paper, the problem of imaging angle correction in the detection of unmanned aerial bridge is studied. It is proposed that the three-point laser is used to obtain the three-point distance, the average object distance is derived, the relative angle of the measured plane is derived, and the pixel resolution obtained by the object distance method is corrected. According to the characteristics of the bridge detection and imaging of unmanned aerial vehicle, the combined filtering method based on morphology is used to construct the image preprocessing algorithm to enhance the contrast of the image and protect the edge information. An edge detection algorithm based on Otsu and Otsu is proposed for edge recognition of UAV images. The tiny noise is eliminated by morphological criterion and the complete crack shape is obtained. A normal width recognition algorithm based on least square fitting of fracture center line is proposed for fracture morphology. The width of cross zone of cross fracture is defined by considering the identification and analysis of fracture shape in special cases. The normal crack width with practical engineering significance is obtained, which provides data support for bridge technical condition assessment based on UAV imaging. At the end of this paper, the actual bridge detection and contrast experiment are carried out, and the width recognition results are compared with the data of the third party detection mechanism. The image crack of the unmanned aircraft is identified by the Xiangjiang second Bridge in Xiangtan City. The results show that the imaging crack identification of unmanned aircraft has good accuracy and is superior to the artificial bridge crack detection to form the crack expansion information map, which meets the practical needs of engineering, and has engineering application value and development potential.
【学位授予单位】：湖南科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U446

【参考文献】